docs: query-vector float64 clause and held VectorIndex reads after DROP INDEX lag behind #176 and #166 #186

Two how-to pages lag behind fixes merged this week:

• docs/modules/ROOT/pages/how-to/databases-and-queries.adoc, query-vector section: the list bullet says a float64 array "on the JVM build" also crosses as one buffer and that "no value changes". Since native: float64 arrays keep their doubles; to_json_list epoch millis floor in integer arithmetic #176 a positional float64 array is narrowed to float32 on both builds, which the next bullet states. The clause in the list bullet now reads as if a float64 array keeps its doubles on the JVM build.
• docs/modules/ROOT/pages/how-to/vectors.adoc: since native: drop_type() drops populated graph types; a held VectorIndex re-reads after a drop #166 the reads of a held VectorIndex (get_size(), get_metadata(), get_quantization(), get_stats()) re-read the engine's metadata on both builds: after DROP INDEX they raise an error that names the index, and a re-created index reports its own configuration. The page does not say so.

Found by a check of the docs, README, and examples against #163, #166, #168, #171, #176, #177, #180, #181, #183, and #185; the other PRs need no docs change (#176, #177, and #185 updated the docs themselves).